# Web LLM - WebGPU로 브라우저에서 LLM 가속하여 실행하기

> Clean Markdown view of GeekNews topic #8977. Use the original source for factual precision when an external source URL is present.

## Metadata

- GeekNews HTML: [https://news.hada.io/topic?id=8977](https://news.hada.io/topic?id=8977)
- GeekNews Markdown: [https://news.hada.io/topic/8977.md](https://news.hada.io/topic/8977.md)
- Type: news
- Author: [xguru](https://news.hada.io/@xguru)
- Published: 2023-04-18T10:17:01+09:00
- Updated: 2023-04-18T10:17:01+09:00
- Original source: [github.com/mlc-ai](https://github.com/mlc-ai/web-llm)
- Points: 14
- Comments: 0

## Topic Body

- 서버 없이 브라우저에서 LLaMA, Vicuna 같은 LLM을 WASM을 이용하여 구동  
- WebGPU로 가속 하지만, 별도로 Apache TVM이 지원하는 GPU 백엔드들(CUDA, OpenCL, Vulkan)도 지원   
- MLC(Machine Learning Compilation) 기술을 사용

## Comments



_No public comments on this page._